View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Electronic Publication Information Center The ISME Journal (2010), 1–9 & 2010 International Society for Microbial Ecology All rights reserved 1751-7362/10 $32.00 www.nature.com/ismej ORIGINAL ARTICLE Carbon and nitrogen fluxes associated with the cyanobacterium Aphanizomenon sp. in the Baltic Sea

Helle Ploug1,3, Niculina Musat2,3, Birgit Adam2, Christina L Moraru2, Gaute Lavik2, Tomas Vagner2, Birgitta Bergman1 and Marcel MM Kuypers2 1Department of Botany, Stockholm University, Stockholm, Sweden and 2Department of Nutrient Group, Max Planck Institute for Marine Microbiology, Bremen, Germany

Carbon and nitrogen fluxes in Aphanizomenon sp. colonies in the Baltic Sea were measured using a combination of microsensors, stable isotopes, mass spectrometry, and nanoscale secondary ion mass spectrometry (nanoSIMS). Cell numbers varied between 956 and 33 000 in colonies ranging in volume between 1.4 104 and 230 104 mm3. The high cell content and their productivity resulted in steep O2 gradients at the colony–water interface as measured with an O2 microsensor. Colonies were highly autotrophic communities with few heterotrophic attached to the filaments. 3 1 Volumetric gross photosynthesis in colonies was 78 nmol O2 mm h . Net photosynthesis was 3 1 3 1 64 nmol O2 mm h , and dark respiration was on average 15 nmol O2 mm h or 16% of gross photosynthesis. These volumetric photosynthesis rates belong to the highest measured in aquatic systems. The average cell-specific net carbon-fixation rate was 38 and 40 fmol C cell1 h1 measured by microsensors and by using stable isotopes in combination with mass spectrometry and nanoSIMS, respectively. In light, the net C:N fixation ratio of individual cells was 7.3±3.4. Transfer of

fixed N2 from to vegetative cells was fast, but up to 35% of the gross N2 fixation in light was released as ammonium into the surrounding water. Calculations based on a daily cycle showed

a net C:N fixation ratio of 5.3. Only 16% of the bulk N2 fixation in dark was detected in þ Aphanizomenon sp. Hence, other organisms appeared to dominate N2 fixation and NH4 release during darkness. The ISME Journal advance online publication, 29 April 2010; doi:10.1038/ismej.2010.53 Subject Category: microbial ecosystems impacts Keywords: Aphanizomenon; ; ammonium release; photosynthesis; respiration

Introduction represented by unicellular picocyanobacteria (Stal et al., 1999; Stal and Walsby, 2000). In the absence of Aphanizomenon sp. is a filamentous, colony- other nitrogen sources, these large filamentous forming N2-fixing cyanobacterium, which frequently cover their nitrogen demand through blooms in lakes and brackish waters. It occurs both N2 fixation, and growth is probably limited by the as colonies and single trichomes during their availability of phosphate (Moisander et al., 2003; growth, which often lasts 2–3 months during the Walve and Larsson, 2007). Aphanizomenon sp. summer in the Baltic Sea (Rolff et al., 2007; accumulates inorganic phosphate during the growth Degerholm et al., 2008). The primary productivity season, and blooms appear to terminate when the in the Baltic Sea by filamentous cyanobacteria, internal storage is used up and bulk concentrations which are represented by Aphanizomenon sp., are low in mid-August in the Baltic Sea (Walve and Anabaena sp., and Nodularia spumigena, represents Larsson, 2007). 44% of the community primary production. These N2 fixation by filamentous cyanobacteria, includ- filamentous organisms comprise 20–30% of the ing Aphanizomenon sp., is considered to be a cyanobacterial biomass in the Baltic Sea. The quantitatively important source of nitrogen to the remaining primary productivity and biomass is plankton community during the summer months in the Baltic Sea (Degerholm et al., 2008). Nitrogen budgets of the Baltic Sea have suggested that Correspondence: H Ploug, Department of Botany, Stockholm Aphanizomenon leaks a substantial fraction of its University, Lilla Frescativa¨gen 5, Stockholm SE-10691, Sweden. fixed nitrogen that is used by other members E-mail: [email protected] 3These two authors contributed equally to this work. (for example picocyanobacteria and heterotrophic Received 21 December 2009; revised 13 March 2010; accepted 15 bacteria) in the plankton community (Larsson March 2010 et al., 2001; Stal et al., 2003; Rolff et al., 2007). C- and N-fluxes in Aphanizomenon sp. H Ploug et al 2 Using stable isotopes, it was shown that 5–10% of transferred to an Eppendorf vial with 1.5 ml sur- 15 the N2 fixed by large cyanobacteria (Aphanizomenon rounding water, and fixed with a drop of 5% Lugol’s sp. and Nodularia sp.) in the Baltic Sea is found in solution. The fixed colonies were stored cold at 4 1C smaller organisms represented by the 2–5 mm size until further analysis. Colonies disintegrate in Lugol’s class, for example picocyanobacteria, after 8–10 h solution, and the filaments sediment at the tip of the incubation time (Ohlendieck et al., 2000). Direct Eppendorf vial; 1 ml containing all filaments from measurements of ammonium or dissovled organic each colony was pipetted into a gridded Sedgewick nitrogen (DON) release relative to N2-fixation rates, Rafter counting chamber (Wildlife supply Company). however, have not been performed in field-sampled Numbers and dimensions of dispersed filaments, cyanobacteria from the Baltic Sea. vegetative cells, and frequency (in no Nanoscale secondary ion mass spectrometry filament per length, and in percent of vegetative (nanoSIMS) is a novel technique, which reveals cells) in each colony was measured under an inverted elemental and isotopic surface composition after microscope (Zeiss, Switzerland) equipped with a measurement of isotopic composition at a spatial digital camera (Axio Cam, Zeiss) connected to an resolution of 50 nm. Combining this technique with image analysis program (AxioVision, Zeiss). Total isotope enriched incubations has enabled measure- cumulative filament length per field was measured ments of N2 and C fixation on a single-cell level in until the mean value was stable and the s.d. was mixed populations (Musat et al., 2008). Hence, o2% of the mean value. using this technique we can now ascribe N2 and C Total abundance of Aphanizomenon sp. trichomes fixation to individual cells of different organisms in and cells in bulk water and in concentrated samples the Baltic Sea. To understand small-scale carbon (see below) was measured using the same counting and nitrogen fluxes in Aphanizomenon sp. colonies, procedure as described above. Five replicates, each we combined microsensors and stable isotope comprising a subsample of 1 ml of 50 ml sample enrichment incubations with mass spectrometry as fixed in Lugol were analyzed both for in situ bulk well as with nanoscale secondary ion mass spectro- water as well as concentrated samples. metry (nanoSIMS). Using microsensors, gross and net photosynthesis as well as dark respiration was

measured directly in colonies as a function of Small-scale measurements of O2 fluxes colony size and cell content in field-sampled Freshly sampled colonies were transferred to a Petri Aphanizomenon sp. in the Baltic Sea. Using stable dish coated by a 3 mm thick agar layer (1% w:w) at isotope in combination with mass spectrometry and the bottom and covered by filtered (0.2 mm) water nanoSIMS, we measured carbon and nitrogen fixed from the sampling site. The diffusion coefficients of by Aphanizomenon sp. cells and its release to the gases, ions, and solutes in 1% agar are close to those surrounding water. in (sea) water (Libicki et al., 1988; Revsbech, 1989). Artifacts from a solid boundary, which limits solute exchange between colonies and the surrounding Materials and methods water, was, therefore, avoided by using a Petri dish coated with agar (Ploug and Grossart, 1999). The Sampling Petri dish was placed in a thermostated container at In August 2008, Aphanizomenon colonies were in situ temperature. Oxygen concentrations were sampled in the upper 10 m of the water column at measured at the colony–water interface using a 0 0 station B1 (N 581 48 28, E 171 37 60) in the Clark type oxygen microsensor (Revsbech, 1989) Stockholm archipelago using a plankton net (Hydro- attached to a micromanipulator. The current was bios: 0.5 m diameter, mesh size: 90 mm). Water at 5 m measured by a picoamperemeter (Unisense PA 2000) depth was collected by a water sampler. The salinity connected to a strip-chart recorder (Kipp and was 6 and the temperature was 19 1C. Samples were Zonen). The electrode was calibrated at anoxic immediately brought to the laboratory in which they conditions and at air saturation. Its 90% response were analyzed by microsensors or incubated with time was o1 s and the stirring sensitivity o0.3%. stable isotopes. The oxygen microelectrode tip was 2 mm wide and its position was observed under a dissection micro- scope with an ocular micrometer. The electrode was Microscopy manually approached toward the colony until it All three dimensions of individual colonies were visually touched the surface. Oxygen-concentration measured under a dissection microscope using a gradients were measured at the colony–water inter- calibrated ocular micrometer. Their volume was face at 50 mm step increments. Three replicates of calculated as ellipsoids: Vol ¼ 4/3p a b c, the O2-concentration gradient were measured in where a, b, and c are the half-axes in the three light and in dark, respectively. The light source was dimensions. A microsensor was used to lift and turn a Schott lamp (KL 1500 LCD) equipped with an colonies on an agar plate covered with water from infra-red cutoff filter and calibrated using an LiCOR the sampling site. Each colony with known size was scalar irradiance sensor. The fluxes of oxygen were thereafter collected using a wide bore pipette, measured at steady state of concentration gradients

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 3 at the colony–water interface, and calculated from coupled to a Delta Plus Advantage mass spectro- Fick’s first law of diffusion: meter was used (Montoya et al., 1996). Before analysis, GF/F filters were freeze dried overnight dC and dehydrated in an HCl-smoke environment in a J ¼D ð1Þ desiccator overnight. Round punchouts (Ø 1 cm) of dr the GF/F filters were packed into tin cups and were tightly pressed into a pill-shaped form. These were 2 1 where J is the flux (nmol O2 cm s ), D is the then loaded onto an autosampler flushed with diffusion coefficient of O2 in the bulk water helium, which automatically dropped the packed 5 2 1 (2.02 10 cm s at 6 and 19 1C; Broecker and filters into the combustion furnace. As a standard, Peng, 1974), and dC/dr is the radial-concentration caffeine was used for calibration. 3 gradient of O2 (nmol O2 cm ). Surface area and volume was calculated as for ellipsoids (Mass, 1994). Carbon production rates were calculated NanoSIMS analysis assuming a photosynthetic quotient of 1.2 mol O2: For NanoSIMS analysis of single Aphanizomenon 1molCO2. cells, Gold-Palladium-coated GTTP filters contain- ing chemically fixed water samples were cut with a round stencil (Ø 5 mm) and mounted on a sample Incubations with stable isotopes holder. The analysis was performed using a Nano- Water samples containing 1.10 108 Aphanizo- SIMS 50 l manufactured by Cameca (Gennevilliers, menon sp. cells per liter (±0.03 108 cells l1; s.e.) France). For each individual cell, we recorded were incubated in 250 ml serum bottles by adding simultaneously secondary ion images of 12C, 13C, colonies harvested by the plankton net to bulk water 12C14N, and 12C15N using four electron multipliers. samples. Microscopy revealed that the mean The measurements and the image and data proces- trichome size was 230±95 mm, and most heterocysts sing were performed as described earlier (Musat were observed within the trichomes. Hence, cells et al., 2008). did not appear damaged by the sampling procedure. The 15N/14N and the 13C/12C ratios of individual The serum bottles were closed with rubber stoppers Aphanizomenon cells were determined for experi- through which 13C bicarbonate was injected to a ments with and without (that is control experi- 15 14 final bicarbonate concentration of 1.83 mM (11% ments) labeled substrate. Subsequently, the N/ N labeling) except for three control bottles. Using a gas and 13C/12C enrichment of individual cells was 15 mouse, 2 ml N2 (Sigma, Germany) was injected to calculated by subtracting the average cellular ratios the sample (12% final labeling). The syringe was of Aphanizomenon in the control experiments from flushed several times with Argon gas between the ratios obtained for individual cells from the injections. All bottles were thoroughly shaken labeling experiment. during 5 min to ensure an even distribution of 15 dissolved N2 in the samples. Many colonies remained intact, and microscopy did not reveal cell Ammonium analysis 15 þ damage during this procedure. Three replicates were The net release of NH4 from cyanobacteria was incubated for 0, 3, or 6 h in light at 300 mmol quantified in the filtrate after the incubations with 15 photons m2 s1, as measured using an LiCor scalar N2 during 0, 3, and 6 h in light or darkness using irradiance sensor, or in dark at in situ temperature the method by Warembourg (1993). Samples were (19 1C) in a thermostated room. The incubations analyzed on a GC-IR mass spectrometer (VG Isogas were stopped by filtration of samples onto Limited, Middlewitch, UK). þ pre-combusted GF/F filters (for EA-MS analysis) The NH4 microenvironment of colonies was (Montoya et al., 1996). The filtrates were filled into modeled, knowing the diffusive properties of colo- 12 ml gas tight exetainers and killed by adding nies from O2 measurements, the net carbon fluxes, 100 ml saturated HgCl solution to each exetainer. the C:N fixation ratio, and fraction of fixed nitrogen 2 þ Subsamples for nanoSIMS analysis were fixed with released as NH4 . We used an analytical diffusion 2% paraformaldehyde and washed before filtration model by Ploug et al. (1997), with a diffusion þ 5 2 1 onto Gold-Palladium-coated GTTP filters (pore size, coefficient of NH4 of 1.6 10 cm s at 6 and 0.22 mm; diameter 25 mm; Millipore, Germany). 19 1C (Li and Gregory, 1974).

Results EA-IRMS analysis To determine the amount of 15N gas and 13C The number of cells in the Aphanizomenon sp. bicarbonate incorporated into biomass, water sam- colonies was a significant function of colony volume ples were filtered onto glass fiber filters and (Figure 1). The largest colonies had a volume of 3 analyzed through mass spectrometry using N2 and 0.023 mm with an equivalent spherical diameter CO2 released by flash combustion in excess oxygen of 340 mm, but most colonies were considerably at 1050 1C. An automated elemental analyzer smaller. The number of cells in individual colonies

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 4 35000 250

m) 250

30000 µ 200 )

25000 -1 200 -1 20000 150

15000 150 100 Cells colony 10000

50 (mm filaments colony Cumulative filament length 100 5000

0 0 50 0.000 0.005 0.010 0.015 0.020 0.025 Colony volume (mm3) 0

Figure 1 Number of cells and cumulative filament length (both Radial distance from colony surface ( shown as symbols) in Aphanizomenon colonies as a function of colony volume. The line represents linear regression between the 270280 290 300 310 320 330 340 variables (please see text). µ O2 ( M)

Figure 2 O2 gradients measured at the colony–water interface during dark (closed symbols) and in light at 300 mmol photons varied from 956 to 33 000 for colonies ranging 2 1 4 4 3 m s (open symbols). Each symbol represents the average value between 1.4 10 and 230 10 mm in volume. of three series of measurements with the s.d. of the mean value

The average number of cells per colony was shown as bars. The curve represents the O2-concentration 6 described as Cells ¼ 1.53 10 (Volcol), where col- gradient fitted to the measured values and used for flux ony volume is measured in mm3. Alternatively, calculations. biomass of filamentous cyanobacteria is measured as meter per length (m l1) (Hajdu et al., 2007; Rolff et al., 2007; Degerholm et al., 2008). The cumulative 1.4

length of filaments in individual colonies varied )

-1 1.2 between 7 and 246 mm, and it was described by L ¼ 11 341 (Volcol); where length, L, is expressed in 1.0 mm. The significance level was R2 ¼ 0.61; n ¼ 37;

colony h 0.8 Po0.0001 for both expressions. The mean cell 2 3 volume was 132 mm , and the mean volume fraction 0.6 occupied by cells in colonies was 0.19. Heterocysts were 1.5% of total cells or 2.02±0.95 mm per 0.4 filament independent of colony volume. 0.2 The cell content and their productivity in colonies 0.0 gave rise to significant oxygen-concentration gradi- exchange (nmol O 2 ents at the colony–water interface (Figure 2). Oxy- -0.2

gen concentrations at the colony surface varied up to Net O 65 mM between the light (300 mmol photons m2 s1) -0.4 and the dark exposed colonies (Figure 2). The gradients were measurable up to 200 mm from the 0.0000.005 0.010 0.015 0.020 surface of a 340 mm large colony. Net photosynthesis Colony volume (mm3) 1 1 was 1.1 nmol colony h , and dark respiration was Figure 3 Net photosynthesis rate (open symbols) and dark 1 1 0.3 nmol colony h as calculated from the oxygen respiration rate (closed symbols) measured as a function of colony gradients measured in light and dark, respectively. volume. Each symbol represents the average value of three series of Hence, dark respiration was 22% of gross photo- measurements with the s.d. of the mean value shown as bars. synthesis in this example. The net photosynthesis and dark respiration measured as a function of colony size are shown respectively, independent of colony size. Combining (Figure 3). Net photosynthesis as well as dark these correlations with the biomass distribution respiration increased proportionally to colony (Figure 1), and assuming a cell-specific carbon volume. Net photosynthesis was described by content of 0.11 pg C mm3 (Helcom, 1988) yielded a 2 NP ¼ 64 (Volcol); R ¼ 0.92 (n ¼ 14; Po0.001), and carbon-specific net photosynthesis rate of 1 1 dark respiration was described by R ¼15 (Volcol); 0.029 h and a dark respiration rate of 0.007 h R2 ¼ 0.94 (n ¼ 14; Po0.001), where volume is mea- independent of colony size. Average cell-specific sured as mm3, and NP and R measured as photosynthesis and respiration rates were 42 and 3 1 1 1 1 1 nmol O2 mm h . Thus, volume-specific gross 10 fmol O2 cell h or 38 and 8.3 fmol C cell h , photosynthesis, net photosynthesis, and dark respectively, assuming a photosynthetic quotient of 3 1 respiration was 79, 64, and 15 nmol O2 mm h , 1.2 (Table 1).

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 5 Table 1 Net photosynthesis and dark respiration in Aphanizomenon sp.

Method Net photosynthesis Dark respiration

nmol C m1 h1 fmol C cell1 h1 nmol C m1 h1 fmol C cell1 h1

Microsensora 5.4 38 1.3 8.3 NanoSIMS 40±12 — — aAssuming a photosynthetic quotient of 1.2.

Figure 4 Distribution of 12C14N (a, d, g), and 15N/14N(e, h), and 13C/12C(f, i) in single Aphanizomenon cells as measured by nanoSIMS. a–c: no label, d–f: t ¼ 3h, g–i: t ¼ 6 h. The size bars are 2 mm.

NanoSIMS analysis of colonies and filaments percent labeling are shown in the bulk in Figure 5. incubated with stable isotopes of C and N showed These varied largely between single cells after 6 h. a rapid transfer of 15N from heterocysts to vegetative Less C relative to N was incorporated into hetero- cells. An example is shown in Figure 4. 13C carbon cysts as compared with vegetative cells. In vege- was evenly distributed in vegetative cells after 3 h tative cells, the average cell-specific carbon assimilation incubation time with only low assimilation into rate was 40±12 fmol C cell1 h1, and the average heterocysts. Heterocysts were also less enriched in cell-specific N assimilation into biomass was 15N as compared with vegetative cells. The occur- 4.8±3.2 fmol N cell1 h1 in light. Hence, the average rence of N-storage products, for example cyano- net C:N assimilation ratio into individual vegetative phycin, was detectable after 3 h as regions with cells was 7.3±3.4 in light (Table 2). Dark respiration 15 1 1 locally intense labeling of N (in cyanophycin was 8.3 fmol C cell h and dark net N2-fixation granules). Hence, incorporation of 15N into proteins rate was 0.28±0.15 fmol N cell1 h1. of heterocysts seemed low, and most of the 15N label We tracked the release of fixed 15N by measuring 15 þ was transported to adjacent vegetative cells. the production of NH4 in the surrounding water 13 15 þ Cell-specific assimilation rates of C and N (Table 2). The release of NH4 was 35% of the total derived from nanoSIMS analysis and converted to gross N2 fixation in light. The corresponding value total net C and total net N assimilated by using was 89% in darkness. The C:N ratio of gross

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 6 production (gross photosynthesis:gross N2 fixation) were similar to those reported in cultured þ was, therefore, 6.3 (Table 2). The NH4 microenvir- colonies of Aphanizomenon sp. (Carlton and Pearl, onment of colonies was modeled, knowing the 1989). Volume-specific gross photosynthesis was 3 1 diffusive properties of colonies from O2 measure- 79 nmol O2 mm h independent of colony size. ments, the net carbon fluxes, the C:N fixation ratio, Hence, there was no indication of diffusion limita- þ and fraction of fixed nitrogen released as NH4 . The tion of photosynthesis by CO2 or bicarbonate in þ NH4 -concentration gradients in a colony with O2 these colonies. Volume-specific gross photosynth- gradients similar to the colony shown in Figure 2 are esis was 10- to 80-fold higher than those measured þ shown in Figure 6. The NH4 concentration at the in surface scums dominated by Nodularia sp. in the colony surface was 5.5 mM when the bulk concentra- Baltic Sea (Ploug, 2008) and of other cyanobacterial tion was 0.15 mM. Hence, ammonium was 437-fold surface blooms including Aphanizomenon in inland enriched inside colonies as compared with the bulk lakes (Ibelings and Maberly, 1998). This high rate is water concentration. among the highest reported in aquatic systems, comparable with or higher than those measured in shallow sediments and microbial mats (Jørgensen Discussion et al., 1983; Villbrant et al., 1990; Lassen et al., 1992). This study is the first to combine microsensors with Carbon-specific net photosynthesis rates were mass spectrometry and nanoSIMS techniques to similar to optimal rates reported using the 14C understand small-scale fluxes of carbon and nitro- method in field-sampled Aphanizomenon from the gen in a Baltic Sea cyanobacterium among those Baltic Sea (Walve and Larsson, 2007) as well as in dominating the summer blooms. It was directly cultures of Aphanizomenon isolated from the Baltic shown that Aphanizomenon is highly productive Sea in the same area as that of this study (Degerholm even during the late stage of its bloom and season in et al., 2006). The 14C method provides an estimate of August. High concentrations of biomass in colonies photosynthesis, which is usually considered to and their productivity lead to significant O2 gradi- represent a rate between gross and net photosynth- ents within these colonies and their surrounding esis (that is the net carbon retainment in the cell). water. The O2 gradients measured during light Using microsensors, it was possible to determine net ) -1 70 h -1 60

50

40

30

20

10

Total C assimilation rate (fmol cell 0 0 2 4 6 8 10 12 Total N assimilation rate (fmol N cell-1 h-1)

Figure 5 Cell-specific carbon and N2-fixation rates as measured Figure 6 Ammonium gradient at colony–water interface of by nanoSIMS. Vegetative cells are shown as open symbols, and Aphanizomenon colony modeled from measured O2 fluxes, þ heterocysts as closed symbols. C-, and N2-fixation rates, and NH4 -release rates.

+ 15 Table 2 Gross and net C:N fixation ratio, N2-fixation rate, and net NH4-release rate in Aphanizomenon sp. traced by N

1 1 a + 1 1 Gross C:N fixation ratio Net C:N fixation ratio N2 net fixation rate (fmol N cell h ) NH4-release rate (fmol N cell h )

Light Light Light Dark Light Dark

6.3b 7.3±3.4c 4.8±3.2 0.28±0.15 2.6 2.2

aMeasured by nanoSIMS. b + Estimating from gross photosynthesis and gross N2 fixation (net N2 fixation+NH4 release). c Net C and net N2 fixation in light of the same cells measured by nanoSIMS.

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 7 photosynthesis as well as dark respiration, and in light, however, were close to a C:N ratio of 6.6. hence gross photosynthesis on single colonies of A high release of DOC or respiration of C could be wild Aphanizomenon sp. Light microscopy of DAPI- expected to maintain a lower somatic C:N ratio, stained colonies revealed that heterotrophic bacter- when the C:N fixation ratio is high. Such high DOC ial colonization of cyanobacterial filaments was low release as reflected by a large difference between (B300 cells mm per filament; data not shown), as rates obtained with isotopes, which are incubated also indicated by the observation that dark respira- for 6 h, and those obtained using the microsensor tion, similar to photosynthesis, was proportional to technique, which measures instantaneous photo- colony volume. Respiration was, therefore, mainly synthesis, was not observed in our study and has due to the cyanobacterium investigated. On average, also not been reported elsewhere. Aphanizomenon it comprised 16% of gross photosynthesis, which sp. seldom occurs in the uppermost water column, partly reflects the cost of N2 fixation. Cell-specific in which incident light intensities are highest carbon assimilation rates compared well between (Hajdu et al., 2007). Integration over a day–night rates obtained with microsensor techniques directly cycle of 10 h:14 h at 3–4 m depth (Secchi depth was in colonies and the total C assimilation rates 6 m, only) resulted in a lower net C:N fixation ratio obtained from 13C enrichment incubations. of (10 38–14 8.3):(10 4.8 þ 14 0.28) ¼ 5.1, be- NanoSIMS analysis showed a large variation of cell- cause N2-fixation rate is reduced and carbon is specific net C and N assimilation. This has also been respired during night. observed in natural populations of other phototrophic We are not aware of any reports on endogenous microorganisms (Musat et al., 2008). Within tri- rhythms of C and N2 fixation in Aphanizomenon sp. chomes, we measured a rapid transfer of fixed as reported for Trichodesmium (Berman-Frank et al., nitrogen from heterocysts to vegetative cells also 2001; Wannicke et al., 2009). However, net N2 reported in Anabaena oscillarioides cultures (Popa fixation by Aphanizomenon sp. was 17-fold lower et al., 2007). Generally, the 15N label intensity in the during darkness as compared with that in light. heterocysts was low compared with that of the Combined NanoSIMS and EA-IRMS analysis of 15N vegetative cells. However, a faint label was evident uptake showed that N2 fixation by Aphanizomenon in this cell type already after 3 h, and a more distinct sp. represented 78% and 16% of the bulk N2 fixation 15 15 N label after 6 h. This is particularly evident in the during light and darkness, respectively. Hence, N2 15 þ ‘polar plugs,’ composed of cyanophycin (arginine:as- fixation and the high NH4 release in the bulk partic acid in a 1:1 ratio), which are located at the during darkness can thus partly be ascribed to other vegetative cell junctions. Hence, although some of the N2-fixing organisms. nitrogen fixed is retained in the heterocysts, the newly Our study directly shows that Aphanizomenon synthesized products of N2 fixation is primarily sp. leaks a substantial fraction of its fixed nitrogen as exported to the vegetative cells in which it apparently ammonium to the surrounding water. Leakage of may be directed into three different routs: toward ammonium, however, could partly be due to break- cellular growth, converted into cyanophycin (N age of cells during the incubations and filtration storage), or being released extracellularly as ammo- procedures or because of lysed cells. Microscopic nium to the environment. In Trichodesmium, the flow analysis and nanoSIMS images did not reveal a offixednitrogenintocyanophycinwasalsoshownby significant number of broken cells. A recent study nanoSIMS, and these granules may partly uncouple C using alternative techniques without potential fil- and N2 fixation from overall growth in cyanobacteria tration artifacts has shown a similar high light- (Finzi-Hart et al., 2009). Moreover, ammonium and dependent ammonium release in Nodularia sp. in DON release in Trichodesmium spp. may amount to the Baltic Sea (Ploug and Adam, unpublished). In þ B50% of N2 fixation (Glibert and Bronk, 1994; this study, modeled NH4 concentrations inside Mulholland and Capone, 2001; Wannicke et al., 2009). Aphanizomenon colonies were 37-fold higher than The average net C:N assimilation ratio in single that of the bulk (Figure 6), and such colonies could cells measured by nanoSIMS was 7.3±3.4 in light, potentially be a microenvironment favorable for and a large fraction (35%) of the newly fixed N was growth of heterotrophic bacteria. Many pelagic released as ammonium to the surrounding water. bacteria are motile, show chemotaxis toward aggre- Hence, gross N2 fixation was approximately 1.5-fold gate nutrient sources, and their colonization times higher than net N2 fixation. Gross C:N fixation are in the order of minutes (Kiørboe et al., 2002). estimated from gross photosynthesis and gross N2 However, bacterial colonization of Aphanizomenon fixation was 6.3 (Table 2). In the Baltic Sea, the C:N colonies was low (data not shown). O2 gradients ratio of large (420 mm) cyanobacteria varies between measured during dark were also less steep as 6 and 20 (Degerholm et al., 2008). The net C:N compared with those measured in cultured colonies fixation ratio varies largely between 1 and 40 (Carlton and Pearl, 1989). Lower abundance of þ presumably because of a variable NH4 release by attached bacteria on Aphanizomenon colonies than the large cyanobacteria (Gallon et al., 2002). Net on Nodularia sp. has been observed earlier in the Baltic C- and N-fixation rates by large cyanobacteria are, Sea (Ulla Rasmussen and B Bergman, unpublished). therefore, not always closely coupled in the Baltic The reason as to why Aphanizomenon filaments Sea. Our estimates of gross and net C:N fixation rates are poorly colonized by heterotrophic bacteria is

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 8 unclear, but may be explained by release of per year, our measurements suggest that its potential secondary metabolites with high concentrations net N2 fixation may be in the order of 2 inside colonies similar to the high concentrations 30–45 mmol N2 m per year, and the gross N2 2 of O2 and ammonium during light as shown here. fixation may be at least up to 70 mmol N2 m per þ During summer, Aphanizomenon sp. occurs at year (excluding NH4 release during darkness). 1 average abundances up to 20 ml in the upper 20 m During a 3-year study, N2 fixation by cyanobacteria of the water column with peaks around 40–60 ml1 (420 mm) in the Baltic Sea was recently reported to in the Baltic Sea (Hajdu et al., 2007; Rolff et al., range between 21 and 78 mmol N m2 per year, and it

2007; Degerholm et al., 2008). This abundance may contributed between 64% and 88% to total N2 be represented by 200–300 average-sized colonies fixation measured at the same stations. (Degerholm per liter (Figure 1). Buoyancy in Aphanizomenon is et al., 2008). Cyanobacterial blooms are charac- controlled by the numerous gas vesicles forming terized by high spatial and temporal variability in larger gas vacuoles that may occupy a large propor- the Baltic Sea, and local concentrations of Aphanizo- tion of the vegetative cell volume. In addition, menon sp. can vary considerably, for example from 2 colony formation increases the buoyancy capacity to 60 ml1 during summer (Hajdu et al., 2007; Rolff

over that held by single filaments (Reynolds and et al., 2007; Degerholm et al., 2008). Local N2 Walsby, 1975). This is supported by the observation fixation by large cyanobacteria, therefore, varies that cell fraction volumes in colonies on average are depending on their abundance, physiological con- as high as 0.19 as measured in this study. Using this ditions, and chemical/physical parameters in the buoyancy capacity, colonies can position them- water column. Budget calculations of inorganic selves at favorable light and/or nutrient conditions nitrogen in the Baltic Sea indicate that total 2 in the water column to capture the energy needed N2-fixation rates range between 60 to 140 mmol N m 15 for N2 fixation through photosynthesis during day per year (Larsson et al., 2001). N tracer studies and migrate below the pycnocline at night presum- suggest N2-fixation rates to range between 101 and ably to take up other nutrients, for example orto- 263 mmol N m2 per year (average: 125 mmol N m2 phosphate (Reynolds and Walsby, 1975; Walsby per year) (Wasmund et al., 2001). On the basis of our et al., 1997; Hajdu et al., 2007). study, we can conclude that Aphanizomenon sp. Combining all measurements, we can make a budget contributes significantly to the overall N2 fixation of of the small-scale fluxes of carbon and nitrogen in which a substantial fraction is directly released as þ Aphanizomenon sp. For each mole carbon fixed at NH4 into the surrounding water. Aphanizomenon saturating light intensity, 0.16 was respired and sp. thus seems to be a significant player in large-

0.16 mol N2 was fixed of which 0.05 mol was released scale carbon and nitrogen biogeochemical fluxes in þ as NH4 to the surrounding water. Our study also the Baltic Sea. showed that net C and N assimilation by Aphanizo- menon sp. was 5.4 nmol C m1 filament h1 and 1 1 0.74 nmol N m filament h , respectively, at saturat- Acknowledgements ing light intensities, and 1.3 nmol C m1 filament h1 and 0.04 nmol N m1 filament h1, respectively, in This study was supported by a Marie Curie Fellowship to darkness. Extrapolating these data back to the Baltic HP (AHICA; 219976) and the Max Planck Society. We Sea and assuming an average abundance of Aphanizo- thank the staff at the Marine Science Center (Asko¨ menon sp. of 20 ml1 in the upper 4 m water column Laboratory) and Marcel Gu¨ nter for assistance in the field (Hajdu et al., 2007; Degerholm et al., 2008), these and in the laboratory. Malin Mohlin is thanked for advice rates correspond to 108 nmol C l1 h1 and using the counting chamber. We also thank three anon- 1 1 ymous reviewers for their thoughtful comments to the 15 nmol N l h at saturating light intensities, and manuscript. 26 nmol C l1 h1 and 0.86 nmol N l1 h1 during darkness. Integrating the net C fixation and net N2 fixation by Aphanizomenon sp. in the upper 4 m assuming saturating light intensities during 10 h light and 14 h darkness yields a net C assimilation References 2 rate of 2.9 mmol C m per day, and a net N2 fixation Berman-Frank I, Lundgren P, Chen YB, Kupper H, rate of 0.55 mmol N m2 per day for only Aphanizo- Kolber Z, Bergman B et al. (2001). Segregation of menon sp. In this calculation, dark respiration nitrogen fixation and oxygenic photosynthesis in the consumes approximately 50% of the C fixed during marine cyanobacterium Trichodesmium. Science 294: 1534–1537. daytime, and net N2 fixation during dark comprises Broecker WS, Peng TH. (1974). Gas exchange rates 10% of the total, daily net N2 fixation, only. The estimate of daily net N fixation is close to the between air and sea. Tellus 26: 21–35. 2 Carlton RG, Pearl HW. (1989). Oxygen induced changes in average measured value of integrated N2-fixation morphology of aggregates of Aphanizomenon rate by cyanobacteria (420 mm) occurring at an flos-aquae (Cyanophyceae): implications for nitrogen 1 average abundance of 20 m l in the Baltic Sea fixation potentials. J Phycol 25: 326–333. (Degerholm et al., 2008). Considering that Aphanizo- Degerholm J, Gundersen K, Bergman B, So¨derba¨ck E. menon sp. blooms approximately 60–90 days (2006). Phosphorus-limited growth dynamics in two

The ISME Journal C- and N-fluxes in Aphanizomenon sp. H Ploug et al 9 Baltic Sea cyanobacteria, Nodularia sp. and Aphanizo- Trichodesmium IMS101: implications for growth. menon sp. FEMS Microbiol Ecol 58: 323–332. Limnol Oceanogr 46: 436–443. Degerholm J, Gundersen K, Bergman B, So¨derba¨ck E. Musat N, Halm H, Winterholler B, Hoppe P, Peduzzi S, (2008). Seasonal significance of N2 fixation in coastal Hillion F et al. (2008). A single-cell view on the and offshore waters of the northwestern Baltic Sea. ecophysiology of anaerobic phototrophic bacteria. Mar Ecol Prog Ser 360: 73–84. PNAS 105: 17861–17866. Finzi-Hart JA, Pett-Ridge J, Weber PK, Popa R, Fallon SJ, Ohlendieck U, Stuhr A, Siegmund H. (2000). Nitrogen Gunderson T et al. (2009). Fixation and fate of C and fixation by diazotrophic cyanobacteria in the Baltic N in the cyanobacterium Trichiodesmium using Sea and transfer of the newly fixed nitrogen to nanometer-scale secondary ion mass spectrometry. picoplankton organisms. J Mar Sys 25: 213–219. Proc Natl Acad Sci 106: 6345–6350. Ploug H. (2008). Cyanobacterial aggregates formed by Gallon JR, Evans AM, Jones DA, Albertano P, Congestri R, Aphanizomenon sp. and Nodularia spumigena in the

Bergman B et al. (2002). Maximum rates of N2 fixation Baltic Sea: small-scale fluxes, pH and oxygen micro- and primary production are out of phase in a environments. Limnol Oceanogr 53: 914–921. developing cyanobacterial bloom in the Baltic Sea. Ploug H, Grossart H-P. (1999). Bacterial production and Limnol Oceanogr 47: 1514–1521. respiration in suspended aggregates—a matter of the Glibert PM, Bronk DA. (1994). Release of dissolved incubation method. Aquat Microb Ecol 20: 21–29. organic nitrogen by marine diazotrophic cyanobacter- Ploug H, Ku¨ hl M, Buchholz B, Jørgensen BB. (1997). ia, Trichodesmium spp. Appl Environ Microbiol 60: Anoxic aggregates—an ephemeral phenomenon in the 3996–4000. ocean. Aquat Microbol Ecol 13: 285–294. Hajdu S, Ho¨glander H, Larsson U. (2007). Popa R, Weber PK, Pett-Ridge J, Finzi JA, Fallon SJ, vertical distributions and composition in Baltic Sea Hutcheon ID et al. (2007). Carbon and nitrogenfixation cyanobacterial blooms. Harmful Algae 6: 189–205. and metabolite exchange in and between individual Helcom (Helsinki commission) (1988). Guidelines for the cells of Anabaena oscillarioides. ISME J 1: 354–360. Baltic Monitoring Programme for the Third Stage, Part Revsbech NP. (1989). An oxygen microelectrode with a D. Biological Determinants. Balt Sea Environ Proc 27 guard cathode. Limnol Oceanogr 34: 474–478. D, Baltic Marine Environment Protection Commission: Reynolds CV, Walsby AE. (1975). Water blooms. Biol Rev Helsinki, pp 23–54. 50: 437–481. Ibelings BW, Maberly SC. (1998). Photoinhibition and the Rolff C, Almesjo¨ L, Elmgren R. (2007). Nitrogen fixation availability of inorganic carbon restrict photosynthesis and abundance of the diazotrophic cyanobacterium by surface blooms of cyanobacteria. Limnol Oceanogr Aphanizomenon sp. in the Baltic Proper. Mar Ecol 43: 408–419. Prog Ser 332: 107–118. Jørgensen BB, Revsbech NP, Cohen Y. (1983). Photosynth- Stal LJ, Albertano P, Bergman B, von Bro¨ckel K, Gallon JK, esis and structure of benthic microbial mats: micro- Hayes PK et al. (2003). BASIC: Baltic Sea Cyanobacteria. electrode and s.e.m. studies of four cyanobacterial An investigation of the structure and dynamics of water communities. Limnol Oceanogr 28: 1075–1093. blooms of cyanobacteria in the Baltic Sea-resposes to a Kiørboe T, Grossart H-P, Ploug H, Tang K. (2002). Mechan- changing environment. Cont Shelf Res 23: 1695–1714. isms and rates of bacterial colonization of sinking Stal LJ, Walsby AE. (2000). Photosynthesis and nitrogen aggregates. Appl Environ Microbiol 68(8): 3996–4006. fixation in a cyanobacterial bloom in the Baltic Sea. Larsson U, Hajdu S, Walve J, Elmgren R. (2001). Baltic Sea Eur J Phycol 35: 97–108. nitrogen fixation estimated from the summer increase Stal LJ, Staal M, Villbrandt M. (1999). Nutrient control of in upper mixed layer total nitrogen. Limnol Oceanogr cyanobacterial blooms in the Baltic Sea. Mar Ecol Prog 46: 811–820. Ser 18: 165–173. Lassen C, Ploug H, Jørgensen BB. (1992). Microalgal Villbrant M, Stal LJ, Krumbein WE. (1990). Interactions photosynthesis and spectral scalar irradiance in coast- between nitrogen fixation and oxygenic photosynth- al marine sediments of Limfjorden, Denmark. Limnol esis in a marine cyanobacterial mat. FEMS Microb Ecol Oceanogr 37: 760–772. 74: 59–72. Libicki SB, Salmon PM, Robertson CR. (1988). The effective Walve J, Larsson U. (2007). Blooms of Baltic Sea diffusive permeability of a nonreacting solute in Aphanizomenon sp.(Cyanobacteria) collapse after microbial cell aggregates. Biotechnol Bioeng 32: 68–85. internal phosphorus depletion. Aquat Microb Ecol Li Y-H, Gregory S. (1974). Diffusion of ions in sea water 49: 57–69. and in deep-sea sediments. Geochim Cosmochim Acta Walsby AE, Hayes PK, Boje R, Stal L. (1997). The selective 38: 703–714. advantage of buoyancy provided by gas vesicles for Mass LRM. (1994). On the surface area of an ellipsoid and planktonic cyanobacteria in the Baltic Sea. New Phytol related integrals of elliptic integrals. Comput Appl 136: 407–417.

Math 51: 237–249. Wannicke N, Koch BP, Voss M. (2009). Release of fixed N2 Moisander PH, Steppe TF, Hall NS, Kuparinen J, Pearl and C as dissolved compounds by Trichodesmium HW. (2003). Variability in nitrogen and phosphorus erythreum and Nodularia spumigena under the influ- limitation for Baltic Sea phytoplankton during ence of high light and high nutrient (P). Aquat Microb nitrogen-fixing cynobacteria blooms. Mar Ecol Prog Ecol 57: 175–189. Ser 262: 81–95. Warembourg FR. (1993). Nitrogen isotopes techniques. In: Montoya JP, Voss M, Kahler P, Capone DG. (1996). A Knowles K., Blackburn TH (eds). Nitrogen Fixation in Soil simple, high-precision, high-sensitivity tracer and Plant Systems. Academic: New York. pp 157–180. assay for N2 fixation. Appl Environ Microb 62: Wasmund N, Voss M, Lochte K. (2001). Evidence of 986–993. nitrogen fixation by non-heterocystous cyanobacteria Mulholland MR, Capone DG. (1996). Stoichiometry of in the Baltic Sea and re-calculation of a budget of nitrogen utilization in cultured populations of nitrogen fixation. Mar Ecol Prog Ser 214: 1–14.

The ISME Journal